2
$\begingroup$

I have three questions about a fixed point at infinity.

  1. How can be proved the following result? (if it is true)

    If we have an entire map of degree $d\geq 2$ (i.e. such that the cardinality of the preimage of every point is $d$) such that it has a superattracting fixed point at $\infty$, then it is a polynomial of degree $d$.

  2. Does every entire function with a superattractive fixed point at infinity (of degree $d$) have the form $f(z) = a_dz^d + a_{d+1}z^{d+1} + \dots$ with $a_d\neq 0$?

  3. This is the version of Böttcher's Theorem that I have:

    Let $f(z) = a_nz^n + a_{n+1}z^{n+1} + \dots$, where $n\geq 2$ and $a_n\neq 0$. Then there exists a local holomorphic change of coordinate $\omega = \phi(z)$ which conjugates $f$ to the $n$-th power map $\omega\mapsto \omega^n$ throughout some neighbourhood of $\phi(0) = 0$. Besides, $\phi$ is unique up to multiplication by an $(n-1)$-st root of unity.

    I want to deduce that this theorem is also satisfied when the superattracting fixed point is $\infty$.

Thank you.

$\endgroup$
1
$\begingroup$

That $z_+=f(z)$ has a super-attracting fixed point at infinity can only be defined as that $w_+=g(w)$ with $g(w)=1/f(1/w)$ has a super-attracting fixed point of the same degree at zero. That is equivalent to the existence and non-nullity of the limits $$ \lim_{w\to 0}\frac{g(w)}{w^d}\ne 0\iff \lim_{z\to\infty}\frac{f(z)}{z^d}\ne 0 $$

  1. if $f$ is entire then this kind of polynomial growth requires it to be a polynomial of degree $d$.

  2. No. $\deg f=d$ says the exact opposite. However $g(w)=b_dw^d+b_{d+1}w^{d+1}+\dots$

  3. This in turn means that there is a coordinate change for $|w|<r$ so that $\phi(g(w))=\phi(w)^d$ so that with $\psi(z)=1/\phi(1/z)$ for $|z|>R$ $$\psi(f(z))=\psi(z)^{d}$$ with the same uniqueness properties. This means nothing more than that outside a large enough disk $B(0,R)$ the terms of $f(z)$ besides $a_dz^d$ are negligible, and that in the rescaled equation $cz_+=(a_d/c^{d-1})(cz^d)$ has $d-1$ solutions for $c^{d-1}=a_d$ to find the normal form with leading coefficient $1$.

$\endgroup$
  • $\begingroup$ Thank you @LutzL . I was not acquainted with infinity as a fixed point and now I understand it. Only a little doubt: at the last part, when you write the rescaled equation, do you mean just to consider the change $f(z) \to cf(z/c)$? $\endgroup$ – dudas Apr 2 '18 at 12:58
  • 1
    $\begingroup$ Essentially yes. You can transform $z_+=f(z)$ by inserting $u=cz+m$ to get the transformed recursion $u_+=cf((u-m)/c)+m$ which does not change the qualitative behavior of the recursionm the polynomial nature of $f$ nor the degree and try to fix the parameters $c,m$ so that some of the transformed coefficients have some standard value, $0$ or $1$. $\endgroup$ – Lutz Lehmann Apr 2 '18 at 16:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.